Computing systems that support voice input often contain many different voice commands that a user may need to learn and remember. Multi-turn conversations between a user and a computing system may further increase voice input complexity, since the user needs to learn appropriate voice command responses for each turn of the conversation. Traditional help and tutorial systems lack suitable learning support, either providing superficial information or overwhelming the user with too much information all at once.